Introduction: The aim of this study was to characterize anti-citrullinated peptide antibody (ACPA) serostatus in pre-clinical rheumatoid arthritis (RA) with and without Human Leukocyte Antigen-Shared Epitope (HLA-SE) alleles. Methods: We identified 192 women in the Nurses’ Health Study cohorts with blood samples obtained 4 months to 17 years prior to medical record-confirmed RA diagnosis. Three controls were selected matched on age, cohort, menopausal status and post-menopausal hormone use. Reactivities to 18 ACPAs were measured using a custom BioPlex platform. We used conditional logistic regression to calculate the relative risk (RR) of RA for any ACPA-positive and peptide-specific ACPA-positive and examined RRs by time between blood draw and RA onset. Measures of multiplicative and additive interaction between any ACPA-positive and HLA-SE were calculated. Results: All ACPAs by peptide groups were significantly associated with RA risk, RRs ranged from 4.7 to 11.7. The association between ACPA and RA varied over time with the strongest association in those with blood draw less than 5 years before onset (RR 17.0 [95% CI 5.8 to 53.7]) and no association 10 or more years prior to onset (RR 1.4 [95% CI 0.5 to 4.3]). Individuals with both HLA-SE and any ACPA-positive had the highest risk of RA. HLA-SE-positive RA cases showed reactivity to more ACPA types than HLA-SE negative (χ2 test for trend, P = 0.01). Conclusions: There is increasing ACPA reactivity up to 10 years before RA onset with the strongest association within 5 years of RA onset. The magnitude of the response to ACPAs, in combination with the presence of HLA-SE, is most important for identifying those individuals with the highest risk of RA.